RBA Annual Conference – 1990 Discussion

1. John Piggott^[*]

I will leave aside discussion of the current account deficit, since that is the topic of the next session, and will largely ignore investment. Trimmed in this way, and boiled down to its essentials, the paper argues for four propositions:

1. The gross private saving ratio is the best ratio to use when assessing private saving performance, and, after inflation adjustment, it looks very stable over the last three decades.
2. Consumption and income move together over the life cycle in a manner not explained by the standard life cycle hypothesis. Demographic transition is therefore unlikely to have a very significant impact on saving ratios.
3. Distortions, notably the tax treatment of nominal interest, superannuation, and the aged pension, are important factors in explaining Australia's persistent private deficits; this has led to persistent capital imports, partly because our relatively high inflation rate exacerbates tax-inflation distortions relative to those in other countries.
4. While public saving has been rapidly increasing in recent years, our public investment is much lower than it was, and this must be factored into the equation.

I plan to discuss these propositions in order.

Proposition 1 can be broken into two parts. Part a, that the gross private saving ratio is the best of the four non-public series to examine when assessing private saving performance, can hardly be contested. This is worth emphasising because of the widespread use of the household ratios in public debate. Substitutability between household and business savings is well established, and the (incomplete) offset repeatedly estimated, in Australia's case a decade ago.^[1] Equally the unreliability and non-comparability of depreciation estimates, across countries and through time, is well documented. Dean et al ^[2] and INDECS ^[3] also suggest use of this ratio to assess savings performance. Its use becomes even more imperative when tax changes are considered, as the INDECS team show when they consider the 1987 switch to an imputation-type corporate income tax, along with the top rate split between individuals and companies.

Part b of proposition 1 is also uncontroversial. Edey's calculations are, broadly speaking, supported by those of Anstie, Gray and Pagan.^[4] They are also consistent with estimates reported by Dean et al for a group of OECD countries.

Now let me turn to proposition 2. There is some controversy in the literature concerning the empirical support for the life cycle hypothesis. The authors tend to favour the view that life cycle consumption smoothing is very limited. This is a particularly important point in the long-run context, because it implies that the demographic transition which the Australian economy will experience over the next forty years is not especially important for projections of private saving performance. The authors draw this conclusion. There they point out that if consumption smoothing is very significant, the demographic transition would

“imply a declining level of private saving in the out years of the projection, preceded at some point by a build up in private savings in preparation for that. On the basis of the projections it would seem not unreasonable to expect the savings build up to have begun already, since the working age population ratio is already close to its projected peak. The fact that this has not happened is consistent with the cross sectional data (from the Australian Household Expenditure Survey) discussed earlier, which indicated very little variation in saving rates across age groups. The implication is that on average households expect to reduce consumption after retirement rather than undergo significant dissavings; if such behaviour is maintained, the changing age composition will not have a major impact on aggregate private saving levels.”

About the only concession to the demographic transition-saving link is the remark that “differences in demographic characteristics may be important in explaining broad differences in saving behaviour between countries”.

The authors and I part company in our views on this issue; a number of points can be made. Firstly, whatever the status of the international literature on consumption smoothing [my superficial reading suggests the matter is unresolved], the Australian evidence derived from Household Expenditure Survey (HES) data which the authors educe in support of their contention is unconvincing.

EPAC^[5] also use HES data (albeit in adjusted form) to support the notion of a small impact of demographic transition on saving performance. The present paper is thus the second authoritative document to use these data for this purpose. It is therefore important that their shortcomings in this regard are clearly spelt out. It should be emphasised that the ABS warn against using the HES to make inferences about saving, and point out that their data are not consistent with National Accounts aggregates. So this should in no way be construed as a critique of the HES itself.

Table 1 compares national totals for various aggregates implied by the HES with Australian National Accounts (ANA) estimates. It reveals a number of important discrepancies between the two sets of data. In general, discrepancies can be traced to differences of definition, coverage, timing, and under reporting. For our purposes, the sources of the differences are less important than their implications for derived estimates of saving behaviour.

Firstly, the most important forms of saving for most of the population, superannuation, and owner-occupied housing, are more or less omitted from the HES. No employer superannuation contributions are recorded, while the ANA total for 1984 is 3.6 per cent of household disposable income (line 3). The HES does ask about personal superannuation and life insurance, but the amount recorded is little more than half the ANA estimate (line 15). The ANA Workers Compensation estimate (line 2) is largely saving as well, since it records provision for payouts rather than the payouts themselves, and the former substantially exceed the latter.

The HES does not treat the flow of services from owner-occupied housing as imputed income, although mortgage interest payments are treated as a “current housing cost” (line 6). Thus a major household life cycle saving mechanism in times of inflation – reduction of real owner occupier mortgage debt via inflation – is not clearly identified. Gross real reductions in owner-occupier mortgage debt outstanding probably dwarf the principal repayments recorded in the HES (line 16). Finally, realised capital gains are not recorded as income in the HES unless classified as trading income. The result of all these considerations is that “saving” in the HES is little more than a quarter of the household saving estimate in the National Accounts (line 14).

These different factors are likely to go some distance in explaining why little saving is revealed in the HES. They do not, however, explain why the HES does not evidence dissaving behaviour among the retired.

Two points might be made in this connection. First, about 80 per cent of retired households own their dwelling outright. The HES makes no attempt to impute this consumption expenditure. So consumption of the elderly does not fall by as much as HES data suggest. Secondly, nearly a third of the income of households with heads aged 65 and over is derived from the HES category “Other” (line 8). Assuming HES estimates of expenditure and income are equal, then to the extent that the capital income in this category is derived from fixed-interest assets, dissaving is taking place at the rate of inflation. An 8 per cent inflation would halve the real value of these assets in less than 10 years.

This HES category, like most capital income related categories, reports values far below those required to balance with National Income totals. It might therefore be argued that were “true” income recorded in the HES, the retired might actually be accumulating real assets, rather than the reverse. To this I have no answer, except to suggest that the under-reporting revealed in the HES may not be uniform across age groups. In particular, retired households, who rely relatively heavily on capital income, may have a more accurate perception of its value than working households whose major income flow derives from labour.

I conclude from all this that it is dangerous to make inferences about the extent to which households smooth consumption over the life cycle, from HES or HES-type data. On this basis, at least, it seems fair to claim that the verdict is not yet in and that the HES has been overused as a source of evidence on saving patterns in Australia.

The second point I would like to take up is the authors' contention that the preretirement savings build up associated with the consumption smoothing view of demographic transition might be expected to have manifested itself already in the Australian case. The rationale for this assertion is that the working population ratio is now close to its peak. The saving peak, however, is not necessarily associated with the working population peak. There is a widespread perception, if not compelling evidence, that the bulk of financial saving for retirement takes place after the age of 45. There are two overlapping stories to tell which might explain this pattern. First, consumption smoothing on an equivalence-scale basis suggests such a pattern – the age might be even later nowadays, when children tend to be born later and to stay at home, and perhaps stay in full-time education, for longer.

Alternatively, it is possible to view saving behaviour as being constrained to a rationed sequence of asset types for a substantial part of adult life. It is sensible to accumulate human capital first, since it will depreciate more quickly the longer human investment is postponed. Next comes the purchase of an owner-occupied dwelling – a sensible next choice for tax reasons – the value of which is frequently constrained by credit rationing. A permanent job brings with it superannuation, again rationed in some degree. Given these other commitments, most families have insufficient liquidity to undertake further discretionary saving until the household head is in his or her 40s. The saving implications of both these stories are reinforced by the age earnings profile.

The baby boom in Australia took place between 1947 and 1961. The oldest baby boomers therefore are now in their early forties. On this basis, the build up of retirement saving will not begin to be evident for at least another 5 years, and it will not be until 2006 that the youngest baby boomer turns 45. Incidentally, to my knowledge none of the Australian foreign debt scenarios has built demographic transition into its calculations. In the light of this analysis it would be interesting to see what impact it might have on some projected paths to foreign debt stabilization.

The third point I wish to make is that, if significant demographic differences can explain differences in saving ratios across countries, then they can presumably also explain differences in saving ratios in one country through time. In preparing a paper for the ANU conference on “Savings Policy and Growth” next month, Bateman et al^[6] regress the gross private saving ratio on a variety of demographic and related variables using a pooled time series, cross country sample. The procedure, of course, is very ad hoc, but when the estimated equations are used to project saving rates into the future, the outcome is striking. In the Australian case, the gross private saving ratio hardly moves over the last three decades, but falls by between one third and one half in the two decades following 2010. Thus, the notion that demographics are very influential in determining private saving ratios is entirely consistent with the stability of that ratio in the Australian case over recent history.

Proposition 3 is effectively argued in Section 3(e), where it is correctly suggested that tax policy has important distortionary effects on the allocation of saving across assets. The impact on aggregate saving, however, is not clear cut for most distortions. Following the authors, I would like to focus on two; superannuation tax, and the aged pension.

On superannuation, the argument in the paper appears somewhat confused. On page 125, the authors state that “the net effect of [the 1983 changes to superannuation provisions] is that saving through superannuation is considerably less favourably treated than before 1983”, but go on to suggest that the data in their Table 1, which covers the 80s, leave “little doubt that concessional treatment of superannuation earnings has led to a shift in the allocation of savings in that direction”. It seems to me much more likely that the numbers in Table 1 reflect a broadening of the superannuation net, and in particular the 3 per cent national wage case increase which was awarded in the form of superannuation. This suggests that the stability of the gross private saving ratio has to be attributed to something other than superannuation saving substitution. As the authors themselves note, the spread of superannuation to lower income levels probably amounts to forced saving. The 1984 HES, which the authors use to support a saving substitution hypothesis, would not reflect this award, which was granted in 1986. In any event, the HES does not record any employer superannuation contributions [see my Table 1].

In order to consider systematically the impact of the role of the aged pension and superannuation on aggregate saving, it is necessary to consider them in combination. The aged pension itself is unlikely to discourage saving to a greater extent than is the case in other comparable countries, since the implied replacement rate is the lowest of these countries. The relevant policies are that the pension is means tested, and that it is possible to take superannuation benefits as a lump sum from the age of 55. These provide the basis for “double dipping”.

The effect on aggregate saving of double dipping depends on alternative policy scenarios. If superannuation had to be taken as an annuity, which is the case in the US for example, saving might be expected to increase, because of the implied reduced access to public pensions through the means test, and because of a possible desire to have access to some liquidity in the years immediately following retirement. If, on the other hand, the age limit is increased, this might tend to reduce saving, since one would have longer to save, and would have less time to spend the accumulation.

It should be noted in this connection that the increases in indexed annuity equivalent pension rates associated with postponed withdrawal is dramatic, both because of the impact of compound interest on the accumulation, and because, when finally withdrawn, residual life expectancy is lower. Under simple assumptions, a worker on average weekly earnings who starts contributing at the age of 35 has accumulated sufficient funds at 55 to give him a replacement rate of 25 per cent. If he defers benefits to age 65, the replacement rate is 50 per cent, and if he works until 65, it is 65 per cent (see Bateman et al).

The fourth proposition concerns the decline in public investment through the eighties. This appears to be well supported by the evidence, although Owen Covick has pointed out to me that privatisation and leasing, as treated in the national accounts, mean that the effective decline in public investment may be overstated. Alesina, Gruen and Jones,^[7] report capital investment figures which take account of the first of these accounting anomalies, and find that it is not sufficient to negate the downward trend in public investment. The impact of leasing on measured public investment is not known, however.

Footnotes

University of N.S.W. I have benefited from discussions with colleagues, especially Owen Covick, Fred Gruen, Geoff Kingston and Peter Stemp. [*]

See Covick, O., “The Australian Labour Market”, Australian Bulletin of Labour, 6, 4, September 1980. [1]

Dean, A., Durand, M., Fallon, J. and Hoeller, P. “Savings Trends and Behaviour in OECD Countries”, OECD, Department of Economics and Statistics Working Papers, No. 67, June 1989. [2]

INDECS “State of Play 6”, Allen and Unwin, 1990. [3]

Anstie, R., Gray, M. and Pagan, A., “Inflation and the Consumption Ratio”, in Pagan, A.R. and Trivedi, P.K., (eds.), The Effects of Inflation: Theoretical Issues and Australian Evidence, Canberra, ANU Press, 1983. [4]

EPAC, “Trends in Private Savings”, Council Paper No. 36, December 1988. [5]

Bateman, H., Frisch, J., Kingston, G. and Piggott, J., “Demographics, Retirement Saving, and Superannuation Policy – an Australian Perspective”, paper prepared for the ANU conference on “Savings, Policy and Growth”, July 1990. [6]

Alesina, A., Gruen, D.W.R and Jones, M.T.(1990), “Fiscal Adjustment, The Real Exchange Rate and Australia's External Imbalance”, C.E.P.R. Discussion Paper No. 233, May. [7]

2. Tom Nguyen^[*]

One of the main themes of the paper is that, in a sense, recent widespread concerns over Australia's “undersaving” have been largely unwarranted. In the concluding section, for example, the authors state: “Thus at least on the simplistic criterion of comparison with past standards, Australia does not seem to have developed any problem of undersaving during the 1980s”.

Concerns about the adequacy of national saving have typically been based on the observed downward trend in the household saving ratio since the mid 1970s. As Edey and Britten-Jones (EBJ) point out, however, household saving is very much an imperfect indicator of the overall saving position. The main reason for this appears to be a tendency for household and corporate savings to offset each other, thus leaving total private saving much more stable than either of its two components. EBJ's arguments for preferring the total private sector saving concept (as opposed to the conventional household saving concept) are compelling.

It is reasonable, however, to ask why the logic should not be extended further. As noted by EBJ themselves (in Section 3(c)), it is generally accepted that there is some offset between private and public savings. Moreover, supporters of the Ricardian equivalence proposition would hold that the offset is complete. Thus, a case can be made for greater reliance on the national (rather than private) saving concept. Of course, the more general the concept, the more diffuse its coverage, so that individual compositional changes may become obscure. But this only serves to highlight the need, in a multivariate system, for monitoring more than one variable. The point remains that, if a broad overview of the system is required, one should look toward more, rather than less, aggregative measures.

The title of the paper suggests that its focus would be on overall national saving and investment, rather than on the private components only. Such a national focus would also be in accordance with the widespread interest in the influence of the national investment-saving gap on recent current account deficits.

If the focus is shifted to the national concepts, it can be seen that concerns about undersaving are perhaps not entirely unjustified. EBJ are careful in acknowledging this distinction (e.g. briefly in Sections 1 and 5). Nevertheless, the bulk of their analysis deals with private saving and investment, implying some degree of preoccupation with private-sector decisions and results. Would the policy implications of their analysis have changed much if the focus had been trained instead on the national aggregates? I suspect that some differences may be quite significant.

Consider, for example, the relevant historical comparison. Figure 13 of the paper shows clearly that gross national saving (as a percentage of GDP) generally declined throughout the second half of the 1970s and most of the 1980s, even though gross private saving has been remarkably stable.

Another approach is to contrast Australia's national saving trends with those of comparable countries. Of the OECD countries listed in Table 5, Canada is probably the most relevant for this purpose, because of similarities with Australia in terms of population, size and growth, output and export composition, capital import requirements, etc. As a recent paper by Ian McLean shows,^[1] the national saving ratios of Australia and Canada exhibited remarkably similar trends during the 100 years or so prior to 1970. By contrast, the ratio for the United States followed a quite different pattern over the same period: before 1910 it was consistently higher than the Australian and Canadian ratios, and after 1950 it was consistently lower. In both of those subperiods, the US national saving rate was generally stable, while Australian and Canadian rates tended to rise.

Table 5 in the EBJ paper shows that in the 1960s Australia's gross national saving ratio was some 3 percentage points higher than Canada's. By the 1980s it had fallen to 0.4 percentage points lower. Of course, the above numbers cannot be taken to indicate any accuracy in measuring the true gaps. Nevertheless, the downward trend in the Australian-Canadian differential may well be significant. EBJ also note a similar, although less marked, fall in Australia's saving ratio relative to the OECD average (Figure 17). In my view, this development is worthy of closer examination. For example, it would be useful to compare Australia with other “small” OECD countries, such as Denmark, Sweden, the Netherlands, etc., which may have more in common with Australia than do the larger economies. Comparison with traditional capital importers is likely to be of particular relevance here.

Yet another way to consider the undersaving question is to compare national saving and national investment. In Figure 16, EBJ show the current account deficit, which reflects not only the investment-saving gap but also the national income account's statistical discrepancy. From this, it is clear that national saving fell further and further behind national investment during most of the 1980s.

It thus appears that, prima facie, there may be some cause for concern that Australia's national saving is too low. (Of course, to resolve this question fully would require a far more complex analysis which involves, among other things, intertemporal optimisation and forecasts about the stream of benefits flowing from current investment.) In turn, this implies the need to highlight the role of public saving, as most of the historical decline in national saving was due to a sharp drop in public saving (see Figure 13). In terms of policy, the critical question then becomes: what caused this sharp fall?

Of course, many of the above observations have not escaped EBJ. The point, however, is that their orientation toward private saving and investment has led them to devote most attention to factors which apparently did not play a leading role in determining the overall outcome. Given that the most significant development in the saving and investment arena during the last decade was the dramatic fall in public saving in the first half of the decade, it is rather disappointing that the authors devote less than 5 per cent of the space taken to a discussion of public saving and investment trends (Section 2(d)).

In view of the central role of the decrease in public saving, one might wish for a closer examination of Nevile's analysis which suggests that much of the change in the fiscal position during this period was due simply to cyclical factors. Were there any other episodes in history during which cyclical variations in the budget deficit assumed such proportions? How sensitive are these estimates to particular assumptions and procedures? If a significant part of the decrease in public saving was due to a drop in the structural (i.e. cyclically adjusted) budget surplus, what were the major reasons behind this policy decision? Were they linked to the secular decline in public investment? The answers to these questions are arguably of greater relevance in the present circumstances than some of the issues which are considered at length in the paper and which relate to the determination of private saving and investment.

In the same vein, I find EBJ's stylised explanations of the current account deficit (Section 4(b)) interesting but, curiously, somewhat inconsistent with their summary of the facts (Section 2). Applying Blanchard's model, they suggest that the current account deficit may have arisen for two reasons. First, if Australia's capital-labour ratio is low relative to other countries, its marginal productivity of capital will be relatively high, thus inducing capital imports and, through consumers' anticipation of future gains in output and income, a current account deficit. Second, microeconomic (particularly tax) distortions which encourage current, relative to future, consumption will also tend to induce current account deficits.

Whilst both of these mechanisms are intuitively appealing (and both have been widely discussed in the literature) they would presumably operate more through the decisions of individuals and firms in the private sector than through government decisions. Yet the data (Table 3) indicate a rather decisive role for the public sector in giving rise to the current account deficit, at least up until the late 1980s.

One of the many useful contributions of this paper is that it draws attention to the “virtual collapse” of public saving between the mid 1970s and mid 1980s, in contrast with the surprising stability of private saving. This might even turn out to be one of its most important contributions, if it succeeds in stimulating further research into the factors and processes which determine public investment and saving in Australia, and into their long-term implications for the economy's overall performance.

Footnotes

Griffith University. [*]

I.W. McLean, “Australian Saving Since 1861: A Comparative Perspective”, University of Adelaide, July 1990, pp. 6–10. [1]

3. General Discussion

In response to John Piggott's doubts about the reliability of the Household Expenditure Survey, Malcolm Edey suggested that, even if those data were not ideal, the close correlation between income and consumption over the life cycle was so striking that it was unlikely to be overturned by better data. He also pointed to the evidence produced by Carroll and Summers (1989) which suggests a very similar result for the US.

Tom Nguyen had argued that the use of Blanchard's model to explain current account deficits was unnecessary in the paper, given the strong public sector contributions to the deficit. Edey responded by pointing out that during the past three decades, the private sector has been in deficit by an average of about 3 per cent of GDP. He felt that it was important to explain this observation and that the Blanchard analysis provided a useful explanatory framework.

The authors defended their focus on gross (rather than net) private saving rates. This is an important measurement issue, because conclusions can depend significantly on which measure is used. Although agreeing that net saving is the more relevant concept in principle, the authors argued that these estimates are unreliable because they require estimates of the depreciation of the capital stock; official estimates of the capital stock are suspect because they have implications which don't seem to accord with other evidence. For detailed argument, they referred to their paper.

There was a discussion of the distortion arising from the interaction between the tax system and inflation. Three main arguments were put forward suggesting that the importance of this distortion has been overplayed. First, with deregulation of the financial system, the nominal interest rate has adjusted and reduced the importance of the distortion. Secondly, marginal tax rates have been reduced, again reducing the importance of the distortion. Finally, it was argued that people have changed their behaviour – investing less in interest-bearing assets (which have low after-tax real returns) and more in equity and other non-interest bearing forms of saving. This issue was raised again during discussion of Carmichael's paper, when it was agreed that the freeing-up of interest rates was not a sufficient condition to remove all the consequences of the inflation-tax distortion.

Is the level of national saving in Australia ‘sufficient’?

• Compared with the OECD, our saving rate is about average but we have a higher-than-average rate of investment. Australia has the second fastest population growth in the OECD (after Turkey).
• Who is to judge the appropriate level of private saving? Different people have different ideas about the appropriate level of their own saving, and surely it is up to them.
• There is widespread public concern that our economy doesn't perform as well as Japan, and other fast-growing economies. As economists, we should contribute to the public debate by pointing to the level of Australian private saving as a substantial reason why our economy doesn't grow very fast . Whether one wants to create incentives to increase private saving depends on one's opinion about the role of public policy.
• Given aggregate private saving, if as a community we decide to develop profitable investments like the Northwest Shelf, there is no reason why we shouldn't draw on foreign saving to do so. We have imported both people and capital over the last three decades and there is no reason to stop this process now.
• Perhaps more important than the aggregate level of saving is where that saving goes – i.e. the quality and mix of investment between dwelling and non-dwelling construction, and plant and equipment.

On the issue of the ‘appropriate’ level of public saving, two points were made:

• If it is important to increase national saving, and the private sector doesn't increase its saving, public saving must be higher.
• Examining Figure 14 in the paper, there has clearly been a structural shift in both public saving and in the public borrowing requirement since 1983. But there has been a substantial stock of public debt built up since 1974 which still requires servicing. This strengthens the case for keeping the level of public saving high.

The related issue of declining public investment was also discussed. It was suggested that the fall is overstated because of a change in public sector behaviour; for example, office blocks are now often rented rather than owned. Further, the critical criterion for judging whether public investment is too low is to ask whether there have been projects which have high social rates of return and which have not been undertaken. While one can point to particular examples (e.g. some roads), the general proposition was hard to defend. There are also many examples of wasteful public investment.

There was discussion about the appropriate deflator to use in calculating real interest rates – should one deflate by a consumer price index or by an asset price index? In studying aggregate saving, there was consensus that the relevant deflator is a consumption price deflator – because the relevant relative price is that between present and future consumption of goods. When examining investment, the answer is less clear. If individuals expect future asset inflation, they may invest in assets on the basis of that expectation, even when (CPI deflated) real interest rates are very high. This may have had a lot to do with the asset price boom of the late 1980s.

Several points were made about the public provision of pensions:

• On average in Australia, there is a 45 per cent decline in private income upon retirement, which is higher than the rest of the OECD.
• In terms of its distorting effect on private saving, the structure of the public provision of pensions is important – in particular, the presence of lump-sum payments and poverty traps with high marginal effective tax rates for the old.
• The participation rate for 60–64 year olds showed a long-term decline until about 1983 after which it trended upwards. Hence, public policy changes over the last five years have not caused the low participation rates in this age group.

There was discussion about whether individuals ‘see through the corporate veil’. Many individuals hold shares indirectly through superannuation funds, and may not be fully aware of the changing value of their superannuation entitlements. Mark Britten-Jones made two points on this issue. First, simple regression analysis suggests that consumption does respond to both household disposable income and to corporate retained earnings with a significant coefficient on corporate retained earnings. Secondly, while all consumers are not so sophisticated, the saving of the rich forms a disproportionate part of aggregate private saving, and they have more incentive to be sophisticated about their level of wealth.